Historical bug tracker for the Pacman package manager.
The pacman bug tracker has moved to gitlab:
https://gitlab.archlinux.org/pacman/pacman/-/issues
This tracker remains open for interaction with historical bugs during the transition period. Any new bugs reports will be closed without further action.
The pacman bug tracker has moved to gitlab:
https://gitlab.archlinux.org/pacman/pacman/-/issues
This tracker remains open for interaction with historical bugs during the transition period. Any new bugs reports will be closed without further action.
FS#74730 - makepkg docs: split-packages and architectures
Attached to Project:
Pacman
Opened by Florian Albertz (rlnm) - Friday, 13 May 2022, 17:57 GMT
Last edited by Allan McRae (Allan) - Friday, 13 May 2022, 22:08 GMT
Opened by Florian Albertz (rlnm) - Friday, 13 May 2022, 17:57 GMT
Last edited by Allan McRae (Allan) - Friday, 13 May 2022, 22:08 GMT
|
DetailsHi everyone,
I have recently spend some time digging into how makepkg builds packages for multiple architectures and while doing so I stumbled across an edge-case when overriding the `arch` array in split packages. If I build the attached 'any.PKGBUILD' on x86_64, I'd expect either of the following to happen: * I get an error due to nonsensical or unsupported configuration * It builds the following packages: * `split` for arch `any` * `split-1` for arch `x86_64` * `split-3` for arch `any` Instead it builds all 4 packages, including the one with an explicit arch of `arm64`, which gets build as `x86_64` instead. In addition, if the pkgbase arch is not `any`, the entire build fails if a single split package does not support the target architecture (see 'non-any.PKGBUILD'). Which at least to me feels unintuitive. I understand that all of this is probably simply unsupported behavior as for most of these cases I don't really see a valid use case. But I feel like that fact should still either be documented (why is overriding the arch array in split packages allowed at all?) or enforced by makepkg by throwing an error. Thank you for looking at this issue and if I either missed something, picked the wrong avenue to report this or can do anything else to help with this, please let me know. Greetings, Florian |
This task depends upon
any.PKGBUILD